Re: BUG REPORT: redisplay assertion failure
>>>> "G" == Gunnar Evermann
<ge204(a)eng.cam.ac.uk> writes:
G> Martin Buchholz <martin(a)xemacs.org> writes:
> Fatal error: assertion failed, file
/net/wanchai/export1/xemacs/home/martin/x/ws/dev/src/redisplay.c, line 6799, cur_elt >=
0
>
> This has been reproduced in debug and optimized, mule and latin-1
> xemacs. I cannot (yet) reproduce it in a -vanilla xemacs.
G> This is probably (somehow) related to the redisplay fix for the
G> widening crash (1998-09-11 in the ChangeLog). There were some other
G> reports of this happening but nobody could ever reproduce it. I still
G> hope that the fix is OK and we are just exposing different redisplay
G> bugs this way...
G> Are you using scroll-in-place? That seems to be pretty good at
G> triggering problems in redisplay (which is a Good Thing).
Yes, I am.
> I might work on fixing this, but I'm afraid I won't
return from an
> excursion into redisplay.c for a year or so.
G> Yeah, I know what you mean, but it would certainly be great for XEmacs
G> if *you* spend a year improving/fixing redisplay :-) I really think we
G> should do something about redisplay: either rewrite it or at least
G> generate some helpful documentation.
> #8 0x1ebb10 in redisplay_window (window={...}, skip_selected=0)
at /net/wanchai/export1/xemacs/home/martin/x/ws/dev/src/redisplay.c:5166
G> could you print the values of:
w-> start_at_line_beg
G> beginning_of_line_p(b,startp)
G> in this frame?
G> Currently I don't have any time at all to look into this (I got a
G> deadline for some papers on Friday and still need to prepare the talk
G> for next week :-) Maybe we can talk about it in Japan.
I'll do the crash demo for you in Japan. I think I can reproduce it
at will, albeit in a heavily Martinized xemacs.
Hmmm. Ok, I can now reproduce this in a vanilla xemacs as follows:
Eval:
(progn
(setq split-window-keep-point nil)
(popper-install))
Then:
M-x shell, C-x 1, C-x 2, button 1 in empty window. KABOOM!
Martin